Monomerizer (or SMILES2Seq, #SMILES2FASTA) is a pipeline that converts proteins, peptides and peptidomimetics, represented as SMILES (chemical formulae), into sequences of amino acids and terminal modifications.
For more information, visit our preprint.
To use the output data to finetune our foundation language model for peptidomimetics, visit: GPepT
To run a Monomerizer demo, use the following command:
python3 run_pipeline.py --input_file demo/example_smiles.txt
- By default, results will be saved to the
output/<datetime>directory. The raw directory contains the raw result, and the standard directory contains the sequences after standardizing them to the standard dictionary accepted by GPepT. - Replace
demo/example_smiles.txtwith the path to your input file containing SMILES strings. (The input file must follow the format of the example files in thedemodirectory.)
--output_dir <path>--min_amino_acids <int>: Minimum number of amino acids required for processing. Default is3.--batch_size <int>: Number of SMILES to process in each batch. Default is100.--max_workers <int>: Maximum number of parallel workers. Default is the number of available CPU cores.-draw: Draws output file like this.
The GPepT model enables the generation of novel peptidomimetic sequences.
For more details, please visit: GPepT on Hugging Face
To decode the generated peptidomimetics back to SMILES and analyze molecules generated by GPepT:
-
Format the output sequences according to the required structure shown in the following example file.
-
Run the analysis pipeline with the following command:
python3 GPepT_analysis_pipeline.py --sequence_file PATH/TO/INPUT_FILE
- By default, the demonomerized SMILES and the analysis report will be saved to the
output/<datetime>directory. - The analysis report includes molecular properties such as:
MolWtLogPTPSAFormalChargeRotatableBondsHydrogenDonorsHydrogenAcceptors
To customize or extend the list of descriptors, feel free to edit src/analyse.py.